Codsoft Data Science Internship Project Repository

Welcome to the Codsoft Data Science Internship Project Repository! This repository showcases the data analysis and machine learning projects completed during my internship at Codsoft.

Overview

This repository contains several projects that demonstrate my skills and experience in data science. Each project folder typically includes:

• Jupyter Notebooks: Python code for data preprocessing, exploratory data analysis (EDA), modeling, and evaluation.
• Datasets: Relevant datasets used for analysis and modeling.
• Documentation: Additional documents such as project proposals, reports, or presentations.

Projects

Task_1: TITANIC SURVIVAL PREDICTION

• Objective:Predicting survival outcomes for Titanic passengers using machine learning to analyze factors like age, gender, and class.
• Methods: Employing machine learning algorithms to predict Titanic passenger survival based on demographic and socio-economic features.
• Results: Achieved 89% accuracy in predicting customer churn.
• Data Source: https://www.kaggle.com/datasets/yasserh/titanic-dataset

Task_2: Iris Flower Classification

• Objective:To create a robust machine learning model for accurately classifying iris flowers based on their sepal and petal measurements.

• Methods: Utilizing supervised learning algorithms such as logistic regression, decision trees, and KNN to classify iris flowers based on sepal and petal measurements.

• Results:logistic regression : accuracy for traning data is 0.9333333333333333, accuracy for test data is 0.8888888888888888 . decision trees : accuracy for training data is 1.0, accuracy for test data is 0.8888888888888888. KNN : accuracy for training data is 1.0, accuracy for test data is 1.0

• Data Source: https://www.kaggle.com/datasets/arshid/iris-flower-dataset

  Task_3: Credit Card Fraud Detection

• Objective: The objective of Credit Card Fraud Detection is to accurately identify and prevent fraudulent transactions to protect customers and minimize financial losses.

• Methods: Methods for Credit Card Fraud Detection include supervised and unsupervised machine learning, deep learning, anomaly detection, ensemble methods, behavioral analytics, real-time monitoring, fraud rules, and data preprocessing techniques.

• Results: accuracy for test data is 0.8934010152284264. accuracy for training data is 1.0

• Data Source: https://www.kaggle.com/datasets/mlg-ulb/creditcardfraud/data